Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <995450c5-a4f2-d2d1-8a44-569726eb7231@kernel.dk>
Date:   Thu, 8 Sep 2022 15:19:55 -0600
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux aarch64; rv:102.0) Gecko/20100101
 Thunderbird/102.2.1
Subject: Re: [PATCH] relay: replay_file_open(): NULL ptr deref in CPU hotplugs
Content-Language: en-US
To:     "Yichun Zhang (agentzh)" <yichun@openresty.com>,
        linux-fsdevel@vger.kernel.org
Cc:     Colin Ian King <colin.i.king@gmail.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        linux-kernel@vger.kernel.org
References: <20220908211731.11534-1-yichun@openresty.com>
From:   Jens Axboe <axboe@kernel.dk>
In-Reply-To: <20220908211731.11534-1-yichun@openresty.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 9/8/22 3:17 PM, Yichun Zhang (agentzh) wrote:
> CPU hotplug may introduce race conditions. We reproduced kernel panics
> due to this NULL ptr deref error when doing frequent random CPU hotplugs
> in a KVM guest.
> 
> Signed-off-by: Yichun Zhang (agentzh) <yichun@openresty.com>
> ---
>  kernel/relay.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/kernel/relay.c b/kernel/relay.c
> index 6a611e779e95..2db69fd527d1 100644
> --- a/kernel/relay.c
> +++ b/kernel/relay.c
> @@ -816,6 +816,9 @@ EXPORT_SYMBOL_GPL(relay_flush);
>  static int relay_file_open(struct inode *inode, struct file *filp)
>  {
>  	struct rchan_buf *buf = inode->i_private;
> +	if (unlikely(buf == NULL))
> +		return -ENOENT;
> +
>  	kref_get(&buf->kref);
>  	filp->private_data = buf;

If it can go away before open is called, what prevents it from going
away after that check but before kref_get() is called?

-- 
Jens Axboe